LLM scaling laws tutorial pairs intuition with practice
The creator describes a simplified associative-memory model where scaling can be predicted from simple intuitions, alongside a hands-on environment for learning to use scaling laws.
TLDR
The creator says they presented the tutorial at MLSS Tübingen. They describe three parts: classic papers and recent developments in scaling laws for large language models, an associative-memory toy model intended to build intuition, and a hands-on environment for putting scaling laws into practice.
Combined views
25K
2 Sources, first seen 16d ago